Pulse Plating of Soft High Magnetic Moment Materials-Properties and Application

Transcription

1 Pulse Plating of Soft High Magnetic Moment Materials-Properties and Application Stanko R. Brankovic *,i, Kursat Sendur, Timothy J. Klemmer, Xiaomin Yang and Earl C. Johns * Seagate Technology 151 Waterfront Place, Pittsburgh, PA Abstract In this paper, we present recent results obtained for pulse plating of 1.8 T CoNiFe alloy. As a starting point, a comprehensive model is developed that explains and mathematically describes the phenomenon of pole tip depression observed in the case of direct current deposition. Based on this model and its results the parameters like pulse time, rest time, and magnitude of the pulse current j pp are defined so the optimal conditions for pulse plating are achieved. The duration of the pulse stage is correlated to a parameter we refer to as the transition time, by taking pulse time to be less than or equal to transition time. The rest time was defined as the time when no current is passing thorough the electrode-electrolyte interface and concentration profile on the interface is recovered to an arbitrary portion of the value existing before the current pulse was applied. Using such a defined pulse function, high moment and low coercivity CoNiFe magnetic alloys were obtained and we were able to achieve uniform deposition in the features down to a 50 nm size. Key words: electrodeposition, magnetic materials, pulse plating, pole tip depression, transition time * Electrochemical Society Active Member i Stanko.R.Brankovic@Seagate.com

2 1. Introduction The ever-continuing drive to increase the areal density of magnetic recording inevitably results in the sub-micron scale-down of magnetic recording heads. The magnetic head core materials are usually electrodeposited NiFe/CoFeNi-based alloys 1,. Deposition of these alloys with controlled uniform thickness, microstructure, and composition in ultra small features (<150 nm) is extremely challenging. The experimental results demonstrate that traditional direct current (DC) plating process faces certain limitations if the widths of the photo-resist pattern defining pole tip is reduced to dimensions below 150 nm. The most obvious example is demonstrated through the appearance of pole tip depression (Figure 1). This is caused by a lower deposition rate in the narrow tip features compared to the larger yoke structures. As an alternative to DC plating process, the pulse plating and pulse reverse plating (PP, PRP) 3 have been already in use for quite some time for deposition of magnetic alloys 4,5,6,7,8. These methods are shown to be capable of depositing magnetic materials in templates with sub-100nm dimensions 9 as well as producing electrodeposits with very smooth surfaces 10 and ultra small grain size 11. In this paper, we present the recent results obtained for pulse plating of 1.8 T CoNiFe alloy 1 in sub-100 nm features with high aspect ratio. The challenge was to define parameters like pulse time t on, rest time t off, Figure 1. Pole tip depression after DC plating.. Results and Discussion and magnitude of the pulse current j pulse, so the optimal conditions for plating are achieved and high quality CoNiFe alloy is obtained..1 DC Plating Model-Transport Limitation for Deposition in the Trench Pole tip region Yoke region In Figure 1 is presented a typical shape of the pole tip structure obtained using a conventional DC plating process. A lack of electrodeposit in the central part of the structure is noticeable which is the so called pole tip depression. This phenomenon becomes even more pronounced if the lateral dimensions of the photo resist pattern defining the pole tip reach sub-100 nm range. The model that explains this phenomenon is Figure. Photo-resist pattern defining writing pole tip.

3 based on the assumption that rheological properties of the plating solution on the solution-electrode interface differ locally if the parameters defining these properties are not uniform. The photo-resist walls at very small distances a part (<150nm), represent the regions of very high curvature 13, (Figure ). In these regions photo-resist and organic additives in the plating bath have a significant mutual interaction 14. This results in formation of an additional viscous layer buildup in the trench defining pole tip compared to the rest of the electrode surface where the amount of additives are confined to adsorbed monolayer 15. This effect is obviously the most pronounced at the very center of the trench (high curvature region) where the distance from the yoke structures (low curvature) is significant comparing to the width of the trench (Figure ). A significant buildup of the organic additives in the pole tip region locally increases the viscosity of the electrolyte near the surface and consequently lowers the mobility/diffusivity of the metal ions 16. This means that the diffusion limitation for electrodeposition in the region defining pole tip occurs before such conditions are reached on the rest of the electrode surface (yokes). As a consequence, a nonuniform deposition rate along the longitudinal pole tip axis occurs causing the formation of pole tip depression. In order to define mathematically the assumptions and conditions explained in the previous text we consider the geometry of the trench defining pole tip schematically represented in Figure 3. We assume that deposition conditions in the trench are diffusion limited while deposition in the regions out of the trench is under kinetic control or the transport in Figure 3. Model of the trench defining the pole tip. Trench width, length and height are marked as CD, λ and δ respectively. these regions is not the limiting factor. This assumption defines a situation where the flow of the ions (flux) in the trench occurs from three sides: right (J R ), left (J L ), and from the top (J N ), (Figure 3). Very shortly after the current is applied to the electrode, the steady state conditions are established in the region defining the pole tip. This means that the spatial concentration profile of the ions in the trench participating in the deposition reaction becomes time independent ( C = 0 ), reaching some fixed value. In order to find the actual t concentration profile in the trench it is appropriate to solve Laplace s equation (eq.(1)) with corresponding boundary conditions appropriate to assumption we made. We take the realistic approximation that the concentration along the y direction do not change significantly due to the fact that the width of the trench is much smaller than the lateral and longitudinal dimension of the trench, (CD<<δ, λ). This reduces the problem to twodimensional case. We also assume that for all positions away from the trench the concentration of ions are constant and equal to the solution bulk value while for the positions in the trench their value changes as we approach the interface at z=0, (eq.(3)). At the interface, the ion concentration is taken to be zero. These conditions and Laplace s equation are defined below:

4 C C + x = 0 (1) λ λ for x,and z δ, C = C () and λ λ for x, z δ, x and 0 Figure 4. Concentration field in the trench defining pole tip, C=f(x,z). and λ λ for x,and z = 0, C = 0 (4) We did not make an attempt to solve this problem analytically 17. The problem was solved approximately using commercial finite element software 18. The solution for concentration distribution within the trench was obtained as a two-dimensional concentration field and the result is presented in Figure 4. Result shows that concentration of the ions in the trench is gradually decreasing from the bulk value C (red color) at the geometrical boundaries of the trench, to the minimum value C=0 (dark blue color) at the bottom of the trench (interface, z=0, λ/ x -λ/). The deposition flux for diffusion limited condition in the trench can be defined in terms of the first Fick s law (eq.(5)). J dep = D. (5) In order to quantify this value it is necessary to know representative concentration gradient in the trench along the z direction, C. The derivation of the function C=f(x,z) over z yields the solution that is dependent on two variables, C = f ( x,z ). Considering that the thickness of diffusion layer in this case is the actual height of the trench (δ) the concentration gradient needs to be averaged along z direction within the boundaries of the trench (δ z 0). The representative of the average gradient, is taken as the gradient at z= δ/, avrg z δ knowing that the dimension of z = the trench along the z direction changes form z=δ in the beginning to z=0 on the end of

5 deposition 19. This yields the expression for to be a function of only one avrg variable, x, and has a form of a higher order polynomial, (eq.(6)-eq.(8)). = f ( x,z ) (6) δ for z = const, = f ( x ),and if z = (7) C 4 = avrg = f(x) = A + Bx + Cx +... (8) Now the expression for deposition flux, eq.(5), can be reformulated in terms of the concentration gradient obtained form the model, eq.(8). J 4 = D f ( x ) = D { A + Bx + Cx...} (9) mod el + Equation (9) describing deposition flux can be conveniently used to formulate the expression for time dependent evolution of the electrode-solution interface in z direction, h z (x,t). 1 hz( x,t ) = t J (10) ρ The t and ρ m are deposition time and molar density of depositing material. If we substitute eq.(9) in to eq.(10) then the final expression describing evolution of the interface adopts the form presented by eq.(11). m hz ( x,t ) = t D f ( x ) mod el = t D { A + Bx + Cx +...} ρ ρ m m (11) In the case of 1.8 T CoNiFe alloys 1 the molar density can be taken as 0.15 mol cm -3. For known deposition times, the only parameter that remains unknown is the diffusion coefficient D. In order to determine this value as an integral Figure 5. Fit of the model defined by eq.(11) (curve-t 4 ) to experimental data from Figure 1(red dots). Curves for t n <t 4 represent model prediction of the evolution of deposit-electrolyte interface in the trench during the deposition.

6 representation of the diffusion transport in the trench, we fit our model to the typical data obtained during CoNiFe DC plating experiment having the D as the only fitting parameter (Figure 5). The values of D obtained from the fit strongly depend on the particular deposit profile that we consider for the fit. Typically, these values of D are up to several orders of magnitude lower than the data reported in the literature for diffusion coefficients of Fe +, Ni + and Co + 0. The evaluation of the integral diffusion coefficient D is the main result of the model and it represents the basic parameter that will be used for proper design of the pulse function. The presented model also predicts that the pole tip depression is larger for a longer deposition times. This prediction is consistent with our experimental observations and proves that assumptions and approximations made in previous text were realistic for a given system.. Design of Pulse Function - Pulse Time ( t on ) The successful application of pulse plating method to the practical problems in electroplating practice (pole tip depression in our case) is very often conditioned by proper determination of the parameters defining the pulse function. For the pulse function described in Figure 6, this means the accurate determination of the pulse time t on, rest time t off, and magnitude of the pulse current density j pulse. These parameters are usually defined through the tedious and numerous trial and error experiments. There are also several attempts in literature to offer some generalized mathematical background for appropriate design of these parameters, based on different approximations and models 3,1,,3,4,5 although for practical Figure 6. Pulse Function purpose, each particular problem where the pulse or pulse reverse plating is applied, requires a separate modeling effort. The duration of the pulse stage (t on ) in our case is correlated to a parameter we refer to as transition time, τ, by taking t on to be less than or equal to τ. By doing this we intend to preclude formation of transport limiting condition during the pulse stage. The estimate of τ is based on the Sands equation 6, eq.(1), which treats the problem of unsteady state behavior on electrode surface during the galvanostatic pulse. C - C 00 = j pulse nf t π D 0.5 (1) The transition time τ is defined as a time elapsed from the beginning of the current pulse until the concentration of the metal ions on the electrode-electrolyte interface reaches

7 zero or some steady value. In our case we make the approximation that concentration of the ions at the interface is zero, C 00 = 0, (13) and we take the overall value of the metal ion concentration in the bulk of the electrolyte (C ) to be equal to the sum of the bulk concentrations of each metal ion constituent of the plating bath 1. Ni,Fe,Co C = Ci (14) Reformulating eq.(1) in terms of t for given approximations yields the expression for transition time τ to be; i t πd nf = = 4 γ j Ni,Fe,Co ( C ) i i τ. (15) pulse Here D is taken to be the result obtained from the previously described model fit to experimental data for pole tip deposited in given trench geometry having a corresponding pole tip depression (Figure 1). The γ is the current efficiency of the pulse. By taking the t on to be less than or equal to the transition time τ we obtain a boundary for the estimate of pulse time as; t on πd nf 4 Ni,Fe,Co ( C ) i i γ j pulse (16).3 Defining Pulse Function Rest Time (t off ) The rest time of the pulse (t off ) is defined in terms of the time that is necessary to recover the concentrations of metal ions on the solution-electrode interface to some arbitrary portion of the starting value prior to the current pulse. In order to mathematically define the situation we consider the diffusion problem in one dimension representing the diffusion of ions through the layer of thickness δ from the bulk of the electrolyte to the electrode-solution interface, t = C D. (17) The initial conditions are that for the moment t=0, the concentration of the ions at the interface is zero, while in region z δ they are constant and equal to some bulk value, C, defined by eq.(14).

8 for for t = t = 0, z t', z = 0, = 0, C = 0,and C = C,and z = δ,c = C z = δ,c = C (18) The driving force for diffusion flux in this case is the concentration gradient across the diffusion layer of thickness δ and it is assumed to be linear. for t 0, C C C =, and t = t', = (19) δ δ = The solution of this problem has an exponential form where concentration at the interface asymptotically approaches the value of C for infinitely long time. C C Dt δ = 1 e. (0) C By taking, = and expressing diffusion layer thickness in terms of transition time C τ, δ = Dτ 7, rearrangement of the eq.(0) yields the close form expression for time t off in terms of transition time τ, eq.(1). t o ff = t 3τ (1) The above expression represents the lower limit for the estimate of the rest time in the pulse function defined in Figure Defining the Pulse Function Pulse Current Density (j pulse ). In the previous discussion we defined the pulse time t on in terms of transition time τ, eq.(16). The expression for transition time includes the pulse current as an independent variable. In the case of alloy deposition the change of deposition current as a variable very often produce a significant change in alloy composition. For CoNiFe films, this could have a detrimental effect on magnetic properties. We performed a series of experiments using different pulse current densities in order to investigate their effect on the deposit composition and its magnetic properties. The ratio of the t on /(t on +t off ) was kept constant and it was calculated in accordance with previous discussions. The CoNiFe bath used in this study was previously reported on several occasions 1,8. We investigated a broad range of pulse currents (from 5mAcm - to 40mAcm - ) and measured the compositions of electrodeposited films. The EDX analysis of deposited material showed very weak dependence of composition on the pulse current used (Figure 7). The Co content was found to be constant and independent on pulse current (78±1 at%) while the Ni content showed gradual decrease from 13±1 at% to 8±1 at% with the increase of pulse current. In the same time the Fe atomic percent has increased from 9±1 at% to 14±1 at%.

9 Figure 7. Composition of CoFeNi alloy vs. Pulse Current Density Figure 8. Magnetic moment (Bs) and Coercivity (Hc easy ) for CoFeNi films vs. Pulse Current Density The magnetic properties of the electrodeposited films and their dependence of pulse current were also investigated. The results are shown in Figure 8. As expected from the composition of the CoFeNi alloys, the measured magnetic moments (Bs) of all films were in the range of 1.8T. Some of the films produced by current densities of 15mAcm - and 35mAcm - showed a slightly higher value (>1.8 T). Since the scattering of the data from the measurements created an error bar that is larger then the actual increase of the measured moment in these two cases, we do not intend to emphasize these findings. The measured values of coercivity for pulse plated CoNiFe films were found to be between 1 and 0.4 Oe. There is a clear trend showing that higher current Figure 9. XRD data for CoFeNi alloy densities produce softer CoFeNi films. obtained for different pulse current densities. The XRD analysis (θ scan) of these Data collected at grazing incidence scan films has been performed and an α=3. interesting correlation was found with measured coercivity values. The data show (Figure 9) that crystal structure of plated films is composed of two phases; face centered cubic (FCC) and body centered cubic (BCC). The relative amount of BCC phase is increasing with the higher magnitude of the pulse current used in deposition. It appears that this has an effect on the coercivity of CoFeNi films. They become softer as the

10 relative amount of BCC phase in deposit is increasing and corresponding pulse current densities used for deposition are higher..5 Pulse Plating of CoFeNi in the sub-100nm Features Results The theoretical and practical considerations presented earlier were used for a practical purpose. The goal was to electrodeposit a uniform and smooth CoNiFe alloy in the pole tip structures with sub-100nm width. The estimate of integral diffusivity of the ions in the trench defining the pole tip (DC plating model) was used to design the pulse function following the procedure presented in sections.-.4. The results are presented in Figure 10. The four top poles with different widths, ranging from 90 to 50 nm are presented without any visible tip depression. The surface of the deposit appears to be very smooth, comparable to the ones obtained by DC plating (Figure1). The focused ion beam cross sections of 80 nm and 50 nm pole tips are presented also for comparison. Figure 10. CoNiFe pole tip structures obtained by pulse plating.

11 3. Conclusion The pulse plating method has been demonstrated as a practical solution for deposition of low coercivity high magnetic moment CoNiFe alloys having features with sub-100nm dimensions and high aspect ratio. Based on the concept of transport limiting conditions for deposition in the trenches defining narrow pole tips, a model was developed that explains and mathematically describes the phenomenon of pole tip depression, observed in the case of direct current deposition. Based on this model and its results, the procedure for design of optimal pulse function was developed that leads to successful application of the pulse plating method to deposition of the CoNiFe alloy in the trenches with particular geometry and dimensions. Also, it is demonstrated that a bath chemistry developed for DC plating of CoNiFe alloy 1,8 can be successfully applied for the pulse plating process producing an electrodeposit with excellent magnetic properties.

12 References 1 T. Osaka, Electrochimica Acta, 44, 3885 (1999). T. Osaka, Electrochimica Acta, 45, 3311 (000). 3 J.C. Puippe and F. Leaman, Theory and Practice of Pulse Plating, AESF, (1986). 4 B.N. Popov, K.-M. Yin, and R.E. White, J. Electrochem. Soc., 140, 131 (1993). 5 H. Shultz and M. Pritzker, J. Electrochem. Soc., 145, 033 (1998). 6 D.L. Grimmett, M. Schwartz, and K. Nobe, J. Electrochem. Soc., 140, 973 (1993). 7 N. H. Phan, M. Schwartz, and K. Nobe, J. Electrochem. Soc., 39, 449, (1994). 8 T. Nakanishi, M. Ozaki, H.S. Nam, T. Yokoshima and T. Osaka, J. Electrochem. Soc. 148, C67 (001). 9 M. Sung, G. Zangari, M. Shamsuzzoha and R. M. Metzger, Appl Phys. Lett., 78, 964 (001). 10 P.T. Tang, Electrochimica Acta, 47, 61 (001). 11 H.Natter, and R. Hempelmann, J. Phys. Chem. 100, 1955 (1996). 1 I. Tabakovic, S. Reimer, V. Inturi, P. Jallen, and A. Thayer, J. Electrochem. Soc.,147, 19 (000). 13 Curvature is defined as 1/r where the radius r could be approximated as the width of the pole tip photoresist pattern. 14 The organic additives like saccharin and Sodium Laurel Sulfate used in the CoNiFe plating bath naturally have a stronger interaction with the photo-resist walls than with the electrode surface and their molecules themselves, which creates conditions for very pronounced capillary effect. This effect scales with 1/r (curvature) where r is being defined previosly W. Plieth, Electrochimica Acta, 37, 115 (199). 16 The diffusivity of the ions, according to Stokes-Einstein relation, is inversely proportional to the viscosity of the solution, D = kt. 6πrη 17 In the literature this problem has been treated and approximate analytical solution involves approach using Green s functions. See for example, M. Necati Ozisik, Boundary Value Problems of Heat Conduction, pp 4, Dover Publisher, NY (1968). 18 The problem was solved using ANSYS commercial software package. 19 The averaging of the concentration gradient in the trench, C, could be done on several different ways. The argument that we follow here assuming z= δ/ is based on the fact that during the deposition in the trench there is continuous change of the trench geometry form z=0, to z= δ (moving boundary problem). One can anticipate that during the deposition of film with thickness δ, half of deposition time, the trench dimension in z direction will be larger than δ/ and half of the time will be smaller than δ/. Without going in to deeper mathematical analysis of this problem we take z= δ/ and = to be z= δ avrg appropriate gradient time average approximation of the real situation during the deposition in the trench. 0 K.-M. Yin, J. Electrochem. Soc. 144, 1560 (1997). 1 S.Roy, D. Landolt, J. Appl. Electrochem., 7, 99 (1997). K.-M.Yin, Surf. Coat. Technol., 88, 16 (1996). 3 K.I.Popov, M.D.Maksimovic, and S.S. Djokic, Surf. Technol., 14, 33, (1981). 4 K.I.Popov, M.D. Maksimovic, and R.M. Stevanovic, Surf. Technol.,,155, (1984). 5 K.I.Popov, D.C.Totovski, and M.D.Maksimovic, Surf. Technol., 19, 181 (1983). 6 H. J. Sand, Phil. Mag. 1, 45 (1905). 7 G. Prentice in Electrochemical Engineering Principles, ed. N.R. Amudson, pp 173, Prentice Hall, NJ (1991). 8 I.Tabakovic, V.Inturi and S. Reimer, J. Electrochem. Soc., 149, C18 (00).